From mboxrd@z Thu Jan 1 00:00:00 1970 From: Ryan Mallon Subject: Re: [PATCH 1/2] ASoC: Add BUG() assertion if max98088_get_channel returns -EINVAL Date: Thu, 29 Sep 2011 21:28:39 +1000 Message-ID: <4E845667.3050404@gmail.com> References: <1317218471.8008.3.camel@phoenix> <4E83AA77.9040700@gmail.com> <20110929103413.GK3697@opensource.wolfsonmicro.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <20110929103413.GK3697@opensource.wolfsonmicro.com> Sender: linux-kernel-owner@vger.kernel.org To: Mark Brown Cc: Axel Lin , linux-kernel@vger.kernel.org, Liam Girdwood , alsa-devel@alsa-project.org, Peter Hsiang , Jesse Marroquin List-Id: alsa-devel@alsa-project.org On 29/09/11 20:34, Mark Brown wrote: > On Thu, Sep 29, 2011 at 09:15:03AM +1000, Ryan Mallon wrote: >> On 29/09/11 00:01, Axel Lin wrote: >>> The callers use the return value of max98088_get_channel as array index to >>> access max98088->dai[] array. >>> Add BUG() assertion for out of bound access of max98088->dai[] array. >> BUG() is pretty heavy handed for a driver. Why not fix the problem >> properly in the callers? > There's nothing constructive that any of the callers can do with an > error code - it's a clear bug in something (probably the driver) if we > get called for a bad control. Simply returning an error code isn't > terribly helpful, it's very obscure what's gone wrong and why. We at > least need a log message. Yeah, it can basically only happen if there is a mismatch between the kcontrol definition and the get_channel function in the driver. Would you be happy with adding a: dev_err(codec->dev, "Bad kcontrol channel name\n"); and then returning the error? It doesn't seem worth panicking the whole driver/system for a bug like this. ~Ryan